The Bell System Technical Journal
Volume XXXI May 1952 Number 3
COPYRIGHT 1952, AMERICAN TELEPHONE AND TELEGRAPH COMPANY
Present Status of Transistor Development
By J. A. MORTON
(Manuscript received March 17, 1952)
The invention of the transistor provided a simple, apparently rugged device
that could amplify — an ability in which the vacuum tube had long held a
monopoly. As with most new electron devices, however, a number of extremely
practical limitations had to be overcome before the transistor could be re-
garded as a practical circuit element. In particular: the reproducibility of
units was poor — units intended to be alike were not interchangeable in
circuits; the reliability was poor — in an uncomfortably large fraction of
units made, the characteristics changed suddenly and inexplicably; and the
"designability" was poor — it ivas difficult to make devices to the wide range
of desirable characteristics needed in modern communications functions.
This paper describes the progress that has been made in reducing these
limitations and extending the range of performance and usefulness of tran-
sistors in communications systems. The conclusion is drawn that for some
system functions, •particularly those requiring extreme miniaturization in
space and power as well as reliability with respect to life and ruggedness,
transistors promise important advantages.
INTRODUCTION
When the transistor was announced not quite four years ago, it was
felt that a new departure in communication techniques had come into
view. Here was a mechanically simple device which could perform many
of the amplification functions over which the electron tube had long
held a near monopoly. The device was small, required no heater power,
and was potentially very rugged; moreover, it consisted of materials
which might be expected to last indefinitely long, and it did not appear
to be too complicated to make.
However, as might be expected for a newly invented electron device,
the practical realization of these promises still required the overcoming
411
412 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
of a number of obstacles. While the operation of the first devices was
well understood in a general way, several items were limiting and puz-
zling, for example:
a — Units intended to be alike varied considerably from each other—
the reproducibility was bad.
b — In an uncomfortably large fraction of the exploratory devices,
the properties changed suddenly and inexplicably with time and tem-
perature, whereas other units exhibited extremely stable characteristics
with regard to time — the reliability was poor.
c— It was difficult to use the theory and then existing undeveloped
technology to develop and design devices to a varied range of electrical
characteristics needed for different circuit functions. Performance char-
acteristics were limited with respect to gain, noise figure, frequency range
and power — the designability was poor.
Before the transistor could be regarded as a practical circuit element,
it was necessary to find out the causes of these limitations, to under-
stand the theory and develop the technology further in order to produce
and control more desirable characteristics.
Over the past two years measurable progress has been made in reduc-
ing, but not eliminating, the three listed limitations.
These advances have been obtained through an improved understand-
ing, improved processes and very importantly through improved ger-
manium materials. As a result:
a — t ne beginnings of method have evolved in the use of the theory to
explain and predict the electrical network characteristics of transistors
in terms of physical structure and material properties.
b — It is now possible to evaluate some of the effects and physical
meaning of empirically derived processes and thereby to devise better
methods subject to control. Previously, inhomogeneities in the material
properties masked the dependence of the transistor electrical properties
even on bulk properties (such as resistivity) as well as on processing
effects.
c — As a result, on an exploratory development level, it is now possible
to make transistors in the laboratory to several sets of prescribed char-
acteristics with usable tolerances and satisfactory yields.
d — Such transistors are greatly improved over the old ones in so far
as life and ruggedness are concerned, and some reduction in temperature
dependence has been achieved. However, it is not to be inferred that
all reliability problems are solved.
e— It has become possible in the laboratory to explore experimentally
some of the consequences of the theory with the result that point con-
PRESENT STATUS OF TRANSISTOR DEVELOPMENT 413
tact devices with new ranges of performance are indicated. Even more
importantly, new p-n junction devices have been built in the laboratory
and these junction devices have indicated an extension in several per-
formance characteristics.
f — By having interchangeable and reliable devices with a wider range
of characteristics, it has become possible to carry on exploratory circuit
and system applications on a more realistic basis. Such applications
effort is, in turn, stimulating the development of new devices towards
new characteristics needed by these circuit and system studies.
It is the purpose of the remainder of this paper to give an over all but
brief summary of recent progress made at Bell Telephone Laboratories
in reducing the above-mentioned limitations on reproducibility, relia-
bility and performance. Since a fair number of types of devices are cur-
rently under development, each with different characteristics to be op-
timized, the data will be presented as a sort of montage of characteristics
of several different types of devices. It is not to be inferred that any one
type of transistor combines all of the virtues any more than such a
situation exists in the electron tube art. Moreover, it will be impossible
in a paper of practical length to present complete detailed characteristics
on all or even several of these devices under development; nor would
it be appropriate since most of these data are on devices currently under
development. Rather, what is desired, is a summary of progress across
the board to give the reader an integrated and up-to-date picture of the
current state of transistor electronics.
REPRODUCIBILITY STATUS
Description of Transistors
Before quantitative data comparing the characteristics of past with
present transistors are presented, it will be useful to briefly review
physical descriptions of the various types of transistors to be discussed.
Fig. 1 shows a cutaway view of the now familiar point-contact cartridge
type transistor. All of the early transistors were of this general construc-
tion and the characteristics of a particular one, called the Type A , will
be used as a reference against which to measure results now obtainable
with new types under current development. Fig. 2 is a semi-schematic
picture of the physical operation of such a device. Pressing down upon
the surface of a small die of /(-type germanium are two rectifying metal
electrodes, one labelled E for emitter, the other C for collector. A third
electrode, the base, is a large area ohmic contact to the underside of
the die of germanium. The emitter and collector electrodes obtain their
414 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
■
:
Fig. 1 — The type A transistor structure.
rectifying properties as a result of the p-n barrier (indicated by the
dotted lines) existing at the interface between the n-type bulk material
and small p-type inserts under each point. When the collector is biased
with a moderately large negative voltage (in the reverse direction) so
that the collector barrier has relatively high impedance, a small amount
of reverse current flows from the collector to the base in the form of
electrons as indicated by the small black circles. Now, if the emitter is
biased a few tenths of a volt positively in the forward direction, a cur-
rent of holes (indicated by the small open circles) is injected from the
Fig. 2— Schematic diagram of a point-contact transistor.
PRESENT STATUS OP TRANSISTOR DEVELOPMENT
415
Fig. 3 — The Ml 689 poiut-contact transistor is typical of those used in minia-
ture packaged circuit functions.
emitter region into the w-type material. These holes are swept along to
the collector under the influence of the field initially set up by the
original collector electron current — thus adding a controlled increment
of collector current. Because of their positive charge these holes can
lower the potential barrier to electron flow from collector to base and
thus allow several electrons to flow hi the collector circuit for every
hole entering the collector barrier region. This ratio of collector current
change to emitter current change for fixed collector voltage is called
alpha, the current gain. In point-contact transistors alpha may be
larger than unity. Since the collector current flows through a high im-
pedance when the emitter current is injected through a low impedance,
voltage amplification is obtained as well.
Some of the new transistors are point-contact transistors similar in
physical appearance to the type A. However, their electrical character-
istics will be shown to be significantly improved not over the old type A
only insofar as reproducibility and reliability are concerned, but also
as to range of performance.
For use in miniature packaged circuit functions, the point contact
transistor has been miniaturized to contain only its bare essentials. Fig.
3 is a photograph of a so-called "bead" transistor (compared to a paper
clip for size) and several of the current development types are being
made in tins form.
In Fig. 4 is shown the famity of static characteristics representative
of the Ml 689 bead type transistor. Note in particular the collector
416
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
V
A
\\
\\
Iv
A N
\\
v\
\ \
s k
t]
\
\
oN
^
V
d \
\
\ \ \ °-^
\ \ °\ "*
\
\
ft
3\
\
\
^
11 \
111
o
<£ ii
UJ <C Q
:
J
!
o
II
w
l-l
q
q
^
o
a
o
lis
< 10
I
u
<*]
q~
\
q
s ^
P c
J3
aavnoA a3iim3
39Vi"IOA H0J.03TI00
l'RKSKXT STATUS OP TRANSISTOR DEVELOPMENT
11:
family which gives the dependence of collector voltage upon collector
current with emitter current as parameter. These characteristics may
he thought of as the dual lo the plate family of a triode.- The slope
of these curves is very nearly the small-signal ae collector impedance of
the transistor.* For a fixed collector voltage of —20 volts, when the
emitter current is changed from zero to one milliampere, note that the
collector current correspondingly changes slightly more than two
milliamperes, indicating a current gain, alpha, of slightly more than
two.
Newest memher of the transistor family recently described by Shock-
ley, Sparks, Teal, Wallace and Pietenpol is the n-p-n junction tran-
sistor. 3, 4 Fig. 5 is a schematic diagram of such a structure. In the center
of a bar of single crystal //-type germanium there is formed a thin layer
Sbase
SINGLE- CRYSTAL
'GERMANIUM BAR
f
p-TYPE
Fig. f) — The n-p-n junction transistor
COLLECTOR
PRIMARY EMITTER CONTROLLED
ELECTRON CURRENT
COLLECTOR
JUNCTION
Bj
SMALL RESIDUAL
COLLECTOR
REVERSE CURRENT
[NOT CONTROLLED BY
EMITTER)
Fig. G — Schematic diagram of a junction transistor.
* As shown by Ryder and Kircher,' the ac collector impedance, r c = R K — R 12 ,
where R*j is the open-circuited output impedance and R !2 is the open-circuit feed-
hack impedance. Usually, R« ;» R^.
418 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
of p-type germanium as part of the same single crystal. Oniric non-
rectifying contacts are securely fastened to the three regions as shown,
one being labelled emitter, one base and one collector. In many simple
respects, except for change in conductivity type from p-n-p in the point-
contact (see Fig. 2) to n-p-n in the junction type, the essential behavior
is similar.
As shown in Fig. 6, if the collector junction is biased in the reverse
direction, i.e., electrode C biased positively with respect to electrode B,
only a small residual back current of holes and electrons will diffuse
across the collector barrier as indicated. However, unlike the point-
contact device, this reverse current will be very much smaller and rela-
tively independent of the collector voltage because the reverse impedance
of such bulk barriers is so many times higher than that of the barriers
produced near the surface in point-contact transistors. Now again, if
the emitter barrier is biased in the forward direction, a few tenths of a
volt negative with respect to the base is adequate, then a relatively large
forward current of electrons will diffuse from the electron-rich n-type
emitter body across the reduced emitter barrier into the base region. If
the base region is adequately thin so that the injected electrons do not
recombine in the p-type base region (either in bulk or on the surface),
practically all of the injected emitter current can diffuse to the collector
barrier; there they are swept through the collector barrier field and
collected as an increment of controlled collector current. Hence, again,
since the electrons were injected through the low forward impedance
and collected through the very high reverse impedance of bulk type p-n
barriers, veiy high voltage amplification will result. No current gain is
possible in such a simple bulk structure and the maximum attainable
value of alpha is unity. However, because the bulk barriers are so much
better rectifiers than the point surface barriers, the ratio of collector
reverse impedance to emitter forward impedance is many times greater,
more than enough to offset the point-contact higher alpha; thus, the
junction unit may have much larger gain per stage. 1 ' 3 ' Fig. 7 is a photo-
graph of a developmental model of such a junction transistor called
the M1752.
The upper part of Fig. 8 is a collector family of static characteristics
for the M1752 n-p-n junction transistor. By way of comparison to those
of the point contact family, note the much higher reverse impedance
of the collector barrier (relatively independent of collector voltage) and
the correspondingly smaller collector currents when the emitter current
is zero. In fact, Fig. 9 is an expanded plot of the lower left rectangle of
the collector family of Fig. 8. The almost ideal straight-line character
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
419
Fig. 7— The Ml 752 junction transistor.
and regular spacing of these curves persists down to voltages as low
as 0.1 volt and currents of a few microamperes. Thus, essentially linear
Class A amplification is possible for as little collector power as a few
microwatts. Constant collector power dissipation curves of 10, 50 and
100 microwatts are shown dotted for reference.
Reproducibility of Linear Characteristics
In describing progress in the reproducibility of those transistor char-
acteristics pertinent to small-signal linear applications, one possible
method is to give the statistical averages and dispersions in the linear
open-circuit impedances of the transistor as defined by Messrs. Ryder
and Kircher. 1 Such a procedure, of course, implies a state of statistical
control in the processes leading to a reasonably well behaved normal dis-
tribution for which averages and control limits can be defined. This
situation can be said to be in effect for most transistors under current
development.
However, for the old type A unit, control simply was not in evidence;
so that in quoting figures on type A's, ranges for commensurate fractions
of the total family will be given. In order that symbols and terminology
4*20
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
will be clear, it will be useful to review briefly the method of defining
the linear characteristics of all transistors. In Fig. 10 is shown a gener-
alized network representing the transistor in which the input terminals
are emitter-base and the output terminals are collector-base. Then, over
a sufficiently small region of the static characteristics, the linear re-
lations between the incremental emitter and collector voltages and
currents may be represented by the pair of linear equations shown. 1
OMA.
L.5
\
-1.0
\
-1.5
\
-2.0
V
-2.5
V
-3.0
\
-3.5
-4.01
-4.51
-5.0,
\
V
\
0.15
Ie =
MA.
K
-0.5
j
-1.0
/
1
1
-1.5
l
-,o,
-2.5.
'
/
f
/
/
/
-,oj
■"/
-4.0.
-4.5
0.25
'
/
/
-y
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
I c IN MILLIAMPERES
Fig. 8 — Static characteristics of the M1752 junction transistor.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
421
0.2
Q-
<
at
o
II
\
\
\
\
I?
\
\
o
tvi
i
o
1
1
\
\
\
\
\
\
\
\
\
\
\
N
//WATT
»
\
N
^
\
\
Kso
"
\
•-
to
//WATI
J
J
"•-jL
h)
-4
4
-/
:=iii
r-r.r""
z=: rr—
""
0.12
I e =0#AMP
0.10
0.08
0.06
O 0.04
/
f
J
3
" 0.02
S
V
'
-0.02
/
-25
'
,-50
'
y-75
-100
-125
-0.04
-150
-0.06
-0.08
.
'-175
20 40 60 80 100 120 140 160 180 200
I c IN MICROAMPERES
Fig. 9 — Expanded plot of the microwatt region of the static characteristics of
the M1752 transistor.
422
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
The coefficients are simply the open-circuit driving point and trans-
fer impedances of the transistor, or the slopes of the appropriate
static characteristics at fixed dc operating currents. These equations
may be represented by any one of a large number of equivalent cir-
cuits of which the one shown in Fig. 11 is perhaps currently most useful.
In this circuit r e is very nearly the ac forward impedance of the emit-
ter barrier, r e is very nearly the ac reverse impedance of the collector
barrier, r b is the feedback impedance of the bulk germanium common
to both, and a is the circuit current gain representing carrier collec-
tion and multiplication if any. It turns out this is very nearly equal
to the current multiplication factor a of the collector barrier mentioned
before. Average values of these elements for the type A transistor are
given in Fig. 11. In Fig. 12 are given the ranges of these parameters
for the type A as of September, 1949, and the control limits* for the
same characteristics for new point-contact transistors now under de-
velopment. For September, 1949, the ranges are taken about the average
values shown in Fig. 11 for the type A transistor. The control limits
given for the present situation apply to a number of different types of
point contact transistors so that the present average values of these
le t
«<
v.t
N
V c = L £ Z„ + L C Z |2
V c = t £ Z 2| + L C Z 22
Fig. 10 — The general linear transistor.
equivalent circuit elements depend upon the type of transistor con-
sidered. In Fig. 13 are given the average values of the characteristics
of the M1729 point-contact video amplifier transistor which bears the
closest resemblance to the older type A transistor. By way of contrast
are given some typical values of the elements for the M1752 junction
transistor which is not yet far enough along in its development to have
design centers fixed nor reliable dispersion figures available.
As Ryder and Kircher have shown, 1 transistors in the grounded-base
connection may be short-circuit unstable if a > 1 and n is too large,
* A.S.T.M. Manual, "Quality Control of Materials," Jan. 1951, Part III, pp. 55-
114.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
423
since r& appears as a positive feedback element. The curve in Fig. 14 is
a plot of the short-circuit stability contour when r e and r c have the
nominal values of 700 and 20,000 ohms. Transistors having a and r b
sufficiently large to place their representative points above this contour
will be short-circuit unstable, i.e., they will oscillate when short-cir-
cuited. Those having an a — r b point below the stability contour will
be unconditionally stable under any termination conditions. The large
unshaded rectangle bounds those values of a and r b , which were repre-
T f = 250 OHMS r c = 20,000 OHMS
?^ = 250 OHMS a = 2
Fig. 11 — Equivalent circuit and average element values of the type A transistor.
ELEMENT
RANGE
SEPTEMBER 1949
RANGE
JANUARY 1952
a
4 : i
± 20 °/o
r c
7 : 1
±30%
r e
3 : i
±20%
r b
7 : i
±25%
Fig. 12 — Reproducibility of point-contact linear characteristics.
TYPE
M 1729
M 1752
r €
120
25
r b
75
250
r c
15,000
5 X10 6
a 2.5
0.95
Fig. 13 — Average characteristics of the M1729 and typical characteristics of
the M1752 transistors.
424
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
sentative of the type A transistor in September, 1949. It is apparent that
the circuit user of type A units had approximately a 50 per cent chance
of obtaining a short-circuit unstable unit from a large family of type A
units. The smaller shaded rectangle bounds the values of a and r b now
realized in the Ml 729 transistor presently under development. Not only
has the spread in characteristics been greatly reduced as shown, but also
the design centers have been moved to a region for which all members
of the M1729 family are unconditionally stable.
It is of interest to note that spreads of the order of ±20 to ±25 per
cent are of the same magnitude as those dispersions now existing amongst
the characteristics of presently available well-controlled electron tubes.
These kinds of data on reproducibility of the linear equivalent circuit
element values hold for practically all classes of point-contact devices
200 300 400 500 600 700
r b IN OHMS
Fig. 14 — Stability contour and ranges of a and r b .
now under development for cw transmission service. While it is too
early to prove that such a situation pertains as well to junction tran-
sistors, there is every reason to expect similar results after a suitable
development period.
Reproducibility of Large-Signal Characteristics for Pulse Application
When electron devices are employed for large-signal applications,
particularly those of switching and computing, it is well known that the
characteristics must be controlled over a very broad range of variables
from cutoff to saturation. In September, 1949, very little attempt was
made to control such pulse use characteristics. In the intervening time,
transistor circuit studies have proceeded to the point where it is possible
to define certain necessary large scale transistor characteristics which,
if met, permit such transistors to be used interchangeably and repro-
ducibly in a variety of pulse circuit functions such as binary counters,
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
425
bit registers, regenerative pulse amplifiers, pulse delay amplifiers, gated
amplifiers and pulse generators. Moreover, it has been possible to meet
these requirements on a developmental level with good yields in at
least three types of point-contact switching transistors. The scope of
this paper will not permit a detailed accounting of the technical features
of this situation and such an account will be forthcoming in future papers
on these particular studies. However, a brief description of some of the
more important pulse characteristics and their tolerances is certainly
pertinent.
In practically all of the transistor pulse handling circuits examined to
date, one characteristic common to all is the ability of the transistor, by
virtue of its current gain, to present various types of two-state negative
resistance characteristics at any one or all of its pairs of terminals. A
typical simple circuit and corresponding characteristic is shown in Fig.
15 for the emitter-ground terminals when a sufficiently large value of
resistance is inserted in the base to make the circuit unstable. In region
I where the emitter is negative, the input resistance is essentially the
reverse characteristic of the emitter as a simple diode. In region II as
the emitter goes positive, alpha, the current gain rises rapidly above
unity. If R b is sufficiently large and alpha, the current gain, is greater
REGION m
(SATURATION)
Fig. ]5 — Emitter-ground negative resistance circuit and characteristic.
426 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
than unity the emitter to ground voltage will begin to fall because of the
larger collector current increments driving the voltage of the node N
negative more rapidly than the emitter current drop through r e would
normally carry it. This transition point is called the peak point. If
then a(r b + R b ) is sufficiently large, in this sense, the input resistance
may be negative in this region II. When the internal node voltage has
fallen to a value near that of the collector terminal the "valley point"
has been reached. At this point, the emitted hole current has reduced the
collector impedance to a minimum value beyond which a is essentially
zero; the transistor is said to be saturated. From this point on the in-
put impedance again becomes positive and is determined almost entirely
by the base and emitter impedances. By terminating the emitter-
ground terminals in various ways with resistor-capacitor-bias com-
binations, such a network can be made to perform monostable, astable
or bistable functions. Under such conditions, the emitter current and
correspondingly the collector current switch back and forth between
cutoff and saturation values. For example, in Fig. 16 is shown a value of
emitter bias and load resistance such that there are three possible
equilibrium values of emitter current and voltage. It may be shown that
the two intersections in regions I and III are stable whereas that in region
II is unstable. Hence, if the stable equilibrium is originally in I, a small
positive pulse A p applied to the emitter will be enough to switch from
stable point I to stable point II and conversely, -A„ will carry it from
the high current point to the low current point. The circuit designer is
interested in reproducing in a given circuit (with different transistors
of the same type) the following points of the characteristic:
a — The off impedance of the emitter — he desires that this be greater
than a certain minimum.
b — The peak point V ep — he desires that this be smaller than a certain
maximum.
c — The value of the negative resistance — he desires that this be greater
than a certain minimum.
d— The valley point V ee , / cs — he desires that these be greater than
certain minima, and
e — The slope in region III — he desires that this be smaller than a
certain maximum so that he may control it by external means.
It may be shown that these conditions can be satisfied for useful
circuits by specifying certain maximum and minimum boundaries on the
static characteristics. Fig. 17 is an idealized set of input or emitter
characteristics. By specifying a minimum value for the reverse resistance
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
427
in region I, condition (a) above is satisfied. By specifying a maximum
slope in region IT and III, condition (e) is satisfied. Now refer to the
idealized collector family in Fig. 18; by specifying a maximum value
to Vca, it is possible to insure condition (d) and by specifying a minimum
value for r co , condition (b) can be satisfied. Finally, in Fig. 19 by de-
TRIGGER
VOLTAGE
Ap OR A v
T"
(v cv ,i ev )
Fig. 16 — Bistable circuit and characteristics showing trigger voltage requirements.
Fig. 17 — Idealized emitter characteristics — slope = Rn
428
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
manding that alpha, as a function of I e , go through a transition from a
negligible value (at small negative I c ) to a value well in excess of unity
(at a correspondingly small positive value of /,.) and maintain its value
well in excess of unity at large values of L , conditions (b) and (c) can
be met.
In Fig. 20 are given the characteristic specifications which must be
met by the M1689 bead type switching transistor now under develop-
ment. With these kinds of limits, circuit users find it possible to inter-
change such M1689 units in various pulse circuits and obtain overall
circuit behavior reproducible to the order of about ±2 db.
V r =-35V
Fig. 18 — Idealized collector characteristics.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
EMITTER CURRENT, I € , IN MILLIAMPERES
Fig. 19 — Effective alpha characteristic.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
429
TEST
CONDITIONS
MINIMUM
MAXIMUM
r co -OFF COLLECTOR
DC RESISTANCE
V c = - 35 V DC
I e = OMA DC
17,500 OHMS
V CI - ON COLLECTOR
VOLTAGE
I c =-2MA DC
l e - 1 MA DC
-3V DC
V C3 -ON COLLECTOR
VOLTAGE
I c = - 5.5 MA DC
I € = 3 MA DC
-4V DC
OFF EMITTER
RESISTANCE
V c = -10 V DC
50,000 OHMS
ON EMITTER
RESISTANCE R„
V c =-10V DC
I f = 1 MA DC
800 OHMS
ai
V c = - 30 V DC
I € = 1.0 MA DC
1.5
a 2
V c = - 30 V DC
I f = + 0.05 MA DC
2.0
a 3
V c = - 30V DC
I c = - 0.1 MA DC
0.3
R 12 - OPEN CIRCUIT
FEEDBACK RESISTANCE
V c = - 10 V DC
I 6 =+1 MA DC
500 OHMS
R 2 I - OPEN CIRCUIT
FORWARD RESISTANCE
V c = -10 V DC
I e = + 1 MA DC
15,000 OHMS
R 22 - OPEN CIRCUIT
OUTPUT RESISTANCE
V c = - 10 V DC
I f = +IMA DC
10,000 OHMS
Fig. 20 — Tentative characteristics for the Ml 689 switching transistor.
RELIABILITY
FIGURE OF
MERIT
SEPTEMBER
1949
JANUARY
1952
AVERAGE
LIFE
= 10,000 HOURS
> 70,000 HOURS
EQUIVALENT
TEMPERATURE
COEFFICIENT OF T c
-1% PER DEG C
-V<*% PER DEG C
SHOCK
?
> 20,000 G
VIBRATION
?
20-5000 CPS
NEGLIGIBLE TO
100 G
Fig. 21— Reliability status.
RELIABILITY STATUS
Life
Reliability figures of merit are not too well defined for electron tubes
and the same situation certainly holds at present for transistors. How-
ever, insofar as these quantities can be presently defined, Fig. 21 shows
430 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
a comparison between the present status and that in September, 1949.
Estimates of the half-life of a statistical family of devices are at best
arbitrary and necessarily amount to extrapolations of survival curves
assuming that a known survival law will continue to hold.* In Septem-
ber, 1949, life tests on type A units had been in effect some 4000 hours.
With the assumption of an exponential survival law, it was not possible,
on the basis of a 4000 hour test, to estimate the slope sufficiently accu-
rately to warrant a half-life estimate in excess of 10,000 hours. These
same type A units have now run on life test for approximately 20,000
hours. With the more reliable estimate of survival slope now possible,
the half -life is now estimated to be somewhat in excess of 70,000 hours.
It should be emphasized, however, that these are type A units of more
than two years ago made with inferior materials and processes. It is
believed that those units under current development, being made with
new materials and processes, are superior; but, of course, life tests are
only a few thousand hours old. Although these new data are encouraging,
it is still too early to extrapolate the data such a long way.
Temperature Effects
Transistors like other semiconductor devices are more sensitive to
temperature variations than electron tubes. In terms of the linear
equivalent circuit elements, the collector impedance, r , and the current
gain, a are the most sensitive. Over the range from — 40°C to 80°C the
other elements are relatively much less sensitive. For type A transistors
these temperature variations in r c and a are shown in Fig. 22. While
these curves are definitely not linear, an average temperature coefficient
for r c of about — 1 per cent per degree was estimated for the purpose of
easy tabulation and comparison in Fig. 21.
Thus, for the early type A, r c fell off to about 20 to 30 per cent of its
room temperature value when the temperature was raised to +80° O;
at the same time a increased from 20 to 30 per cent over the same
temperature range. Today, this variation has been reduced by a factor
of about four for r c in most point-contact types, the variations in the
current gain being relatively unchanged. Fig. 23 illustrates the tem-
perature dependence of r c and a for the M1729 transistor now under
development. Again, for purposes of easy comparison in Fig. 21, the
actual dependence of Fig. 23 was approximated by a linear variation and
* Estimates of life, of course, depend upon definitions of "death". For these
experiments, the transistors were operated as Class A amplifiers. A transistor is
said to have failed when its Class A gain has fallen 3 db or more below its starting
value.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
4.31
only the slope given in Fig. 21. For linear applications such as the
grounded base amplifier, the Class A power gain is approximately pro-
portional to a 2 r c ; hence the gain of such an amplifier will stay essentially
constant within a db or two over the temperature range from — 40°C
to +80°C. For pulse applications, and of importance to dc biasing with
point-contact transistors, is the fact that the dc collector current (for
fixed emitter current and collector voltage) will change at about the
20 30 40 50 60 70 80 90
TEMPERATURE IN DEGREES CENTIGRADE
Fig. 22 — Collector resistance and a versus temperature for type A transistor
ujI30
»-> 80
CURRENT
GAIN a^
'
COLLECTOR^.
RESISTANCE
20 30 40
TEMPERATURE
50 60 70 80 90
IN DEGREES CENTIGRADE
Fig. 23 — Collector resistance and a versus temperature for type M1729 transistor.
432
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
same rate as does r c , the small signal collector impedance. Similar im-
provements have been made in these variations for switching transistors
and Fig. 24 is a series of graphs showing how the M1689 bead type
switching transistor changes the pulse characteristics defined in Fig. 20
with respect to temperature. For those switching functions examined
to date, it is believed that these data mean reliable operation to as high
as +70°C in most applications and perhaps as high as +80°C in others.
-fe-
-0--
■>
■^>
re
"J so
^
— o
Jjn.
— """1
~ccT
"**o»,
■^
r c
I — o—
^
"^
20 30 40 50 60 70 80 20 30 40 50 60 70 80
TEMPERATURE IN DEGREES CENTIGRADE
Fig. 24 — Temperature behavior of the M1689 transistor.
In junction transistors the laws of temperature variation are not so
well established, the device being in a much earlier stage of development.
Preliminary data indicate smaller variations in the small signal pa-
rameters such as a and r c . On the other hand, variations in the dc cur-
rent, particularly I c0 , are many times greater, of the order of 10 per
cent per degree centigrade.* The only saving grace here is the fact that
Ice is normally very much less than the actual operating value of I c .
In summary, it may be said that while significant improvements have
been made in temperature dependence to the point where many appli-
cations appear feasible, it is not to be inferred that the temperature
limitation is completely overcome. Much more development work of
device, circuit and system nature is required to bring this aspect of
reliable operation to a completely satisfying solution.
Shock and Vibration
With regard to mechanical ruggedness, current point-contact tran-
sistors have been shock tested up to 20,000 g with no change in their
* I, a is the collector current at zero emitter current.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
433
electrical characteristics. Vibration of point-contact and junction tran-
sistors over the frequency range from 20 to 5000 cps at accelerations of
lOOg produces no detectable modulation of any of the transistor elec-
trical characteristics, i.e., such modulation, if it exists, is far below the
inherent noise level. At a few spot frequencies in the audio range, vi-
bration tests up to lOOOg accelerations similarly failed to produce dis-
cernible modulation of the transistor characteristics.
MINIATURIZATION
FIGURE OF MERIT
TYPE A
SEPTEMBER
1949
JANUARY NEW
1952 ' DEVELOPMENT TYPE
VOLUME
'/so in3
'/2000 'N 3
POINT- M1689
V500 IN 3
JUNCTION -MI752
MINIMUM COLLECTOR
VOLTAGE FOR
CLASS A OPERATION
30 V
2V
POINT- M1768, MI734
0.2 V
JUNCTION -M1752
MINIMUM COLLECTOR
POWER FOR
CLASS A OPERATION
50 MW
2MW
POINT- MI768
10// W
JUNCTION -M 1752
CLASS A
EFFICIENCY
20%
35%
POINT -MI768. MI729
49%
JUNCTION-MI752
Fig. 25 — Miniaturization in space and power drain.
MINIATURIZATION STATUS
Space Requirements
In smallness of size, the transistor is entering new fields previously
inaccessible to electron devices. The cartridge structure (see Fig. 25),
such as the type A, has a volume of -£$ cubic inch, compared to about |
cubic inch for a sub-miniature tube and about 1 cubic inch for a minia-
ture tube. Under current development, the Ml 689 bead point-contact
transistor has substantially similar electrical characteristics to the M1698*
cartridge switching unit but occupies only about ^irVir cubic inch. The
M1752 junction bead transistor has a volume of approximately sfa cubic
inch but this may be reduced to the same order as the point-contact bead
if necessary. For further substantial size reductions in equipment, the
next move must comprise the passive components. It should be pointed
out that the low voltages, low power drain, and correspondingly lower
equipment temperatures should make possible further reductions in
passive component size.
* The Ml 698 transistor is a cart ridge type point-contact transistor with elec-
trical characteristics designed for switching and pulse applications. This unit is
proving useful in the laboratory development of new circuits or in cases where
miniature packages arc unnecessary.
434 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
Power Requirements
The transistor, of course, has the inherent advantage of requiring no
heater power; moreover, significant advances have been made in the past
two years in reducing the collector voltage and power required for prac-
tical operation. Consider the minimum collector voltage for which the
small-signal Class A gain is still within 3 to 6 db of its full value. In
September, 1949, the type A transistor could give useful gains at col-
lector voltages as low as 30 volts. Today, several point -contact devices
(M1768 and Ml 734) perform well with collector voltages as low as 2 to
6 volts even for relatively high-frequency operation. One junction tran-
sistor, the M1752, can deliver useful gains at collector voltages as low
as 0.2 to 1.0 volt. Under these same conditions, the minimum collector
power for useful gains may be as low as 2-10 mw for point-contact
devices and as low as 10 to 100 /xw in the case of the junction transistors.*
Class A efficiencies have been raised for the point-contact devices to as
high as 30-35 per cent and for junction transistors this may be as high
as 49 per cent out of a maximum possible 50 per cent. Class B and C
efficiencies are correspondingly close to their theoretical limiting values.
PERFORMANCE STATUS
Exact electrical performance specifications for the transistor depend,
of course, upon the intended applications and the type of transistor
being developed for such an application. These types are beginning to
be specified; and in fact, they are already so numerous that mention of
only a few salient features of some of them will be attempted. Bear in
mind, as was pointed out before, that no one transistor combines all
the virtues any more than does any one tube type. Fig. 26 attempts to
compare the progress made in several important performance merit
figures by development of several point-contact and junction types
during the last two years. Again the reference performance is that of
the type A as of September, 1949.
Some switching and transmission applications need transistors having
high current gain. By going to a point-junction structure, useful values
of alpha as high as 50 are now possible with laboratory models.
For straight transmission applications, the single stage gain of point-
contact types (M1768, M1729) has been increased to 20-24 db, whereas
for the M1752 junction type the single stage gain may be as high as
45-50 db.
* In some special cases, depending upon the application, practical operation
may be obtained for as little as 0.1 to 1.0 microwatt.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
435
PERFORMANCE
FIGURE OF MERIT
TYPE A
SEPTEMBER
1949
JANUARY
1952
NEW
DEVELOPMENT TYPE
a -CURRENT GAIN
5X
SOX
JUNCTION
SINGLE STAGE
CLASS A
GAIN
18 DB
22 DB
POINT- M 1729, MI768
45 DB
JUNCTION -M 1752
NOISE FIGURE
AT 1000 CPS
60 DB
45 DB
POINT- M 1768
10 DB
JUNCTION-M1752
FREQUENCY
RESPONSE
SMC
7-IOMC
POINT- MI729
20-50 MC
POINT- M 1734
CLASS A
POWER OUTPUT
0.5 WATT
2 WATTS
JUNCTION
SWITCHING
CHARACTERISTICS
NONE
GOOD
POINT-M1698,M1689
MI734
FEEDBACK
RESISTANCE
250 OHMS
70 OHMS
POINT-M1729
^fg£ PHOTOCURRENT
UAhm RATIO
2:1
20 :i
JUNCTION-MI740
Fig. 26 — Performance progress.
For high-sensitivity low-noise applications, the point-contact devices
have been improved to have noise figures of only about 40-45 db, whereas
the M1752 ?i-p-n transistor has been shown to have noise figures in the
10-20 db range. All such noise figures are specified at 1000 cps and it
should be remembered that they vary inversely with frequency at the
rate of about 11 db per decade change in frequency.
For video, I.F., and high-speed switching applications, measurable
improvement has been attained in the frequency response. For video
amplifiers up to about 7 mc, the M1729 point-contact transistor is
capable of about 18-20 db gain per stage. For high-frequency oscillators
and microsecond pulse switching, the M1734 point-contact transistor is
under development. Preliminary models of 24 mc I.F. amplifiers using
the M1734 have been constructed in the laboratory, these amplifiers
having a gain of some 18-24 db per stage and a band-width of several
megacycles. However, more work needs to be done on the M1734 to
reduce its feedback resistance. For pulse-handling functions, such M1734
units work very nicely as pulse generators and amplifiers of \ micro-
second pulses, requiring only 6-8 volts of collector voltage and 12-20
mw of collector power per stage. The amplified pulses can have ampli-
436 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1052
fcudes as large as 4-5 volts out of a total collector voltage of volts and
rise times as little as 0.01-0.02 microsecond.
By increasing the thermal dissipation limits of junction transistors,
the Class A power output has been raised to 2 watts in laboratory models.
This, however, does not represent an intrinsic upper limit but rather a
design objective for a particular application.
Characteristics suitable for switching are now available in the M1G98,
Ml 089 and M1734 point-contact types, as previously described, but
this is a continually evolving process and more work certainly remains to
be done. At present it is possible to operate telephone relays requiring
as much as 50 to 100 ma with Ml 089 and M1098 point-contact tran-
sistors.
New junction-type phototransistors 6 represent a marked advance over
the earlier point-contact type. 6 While their quantum efficiencies are not
as high as those of the point-contact types, nevertheless the light/dark
current ratios are greatly improved and the collector impedance has been
raised 10-100 times thus making possible much greater output voltages
for the same light flux.
SOME SELECTED APPLICATIONS
Data Transmission Packages
To determine the feasibility of applying transistors in the form of
miniature packaged circuit functions, several of the major system func-
tions of a pulse code data transmission system have been studied. This
investigation has been undertaken under the auspices of a joint services
engineering contract administered by the Signal Corps.
It was desired that these studies should lead to the feasibility develop-
ment of unitized functional packages combining features of miniaturiza-
tion, reliability and lower power drain. Accordingly, it was necessary
to carry on in an integrated fashion activities in the fields of system,
circuit and device development to achieve these ends. In particular,
circuit and system means have been developed to perform with tran-
sistors the functions of encoding, translation, counting, registering and
serial addition. The M1728 junction diode, M1740 junction photocell
and M1089 bead switching transistor are direct outgrowths of this
program and are the devices used in the circuit packages.
At this point, the major system functions shown in Fig. 27 have been
achieved with interchangeable transistors. These major system functions
are in turn built up of some seven types of smaller functional packages
listed in Fig. 28. The end result of this exploratory development can be
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
437
said to have demonstrated the feasibility of such a data transmission
system in the sense that a workable (though not yet optimal) system
can be synthesized from reproducible transistor-circuit packages which
have been produced at reasonable yields and with reasonable (though
not yet complete) service reliability- Further development work would
be needed in all phases to make such a system of packages suitable for
field use. It is estimated that the present laboratory model requires
about one-tenth the space and power required to do the same job with
present tube art. Fig. 29 is a photograph of a transistor bit-register
package and Fig. 30 is another photograph of such packages showing
both sides of the various types employed.* Actual final packages would
1. 4 DIGIT REVERSIBLE BINARY COUNTER
2. 6 DIGIT ANGULAR POSITION ENCODER
3. 6 DIGIT GRAY-BINARY TRANSLATOR
4. 5 DIGIT SHIFT REGISTER
5. 2 WORD SERIAL ADDER
Fig. 27 — System functions tested.
DEVELOPMENT
PACKAGE
TYPE
PACKAGE FUNCTION
DEVELOPMENT
TRANSISTOR, DIODE
TYPES USED
M 1731-1
REGENERATIVE GATE
M 1689
M 1727
M 1732-1
M 1736
M 1790
BIT REGISTER
M 1689
M 1727
M 1734
M 1733-1
M 1792
PULSE AMPLIFIER
M 1689
M 1735-1
M 1747-1
M 1748-1
M 1751 -J
M 1751-2
M 1751-3
DIODE GATE
M 1727
400 A
M 1745-1
M 1791
BINARY COUNTER
M1689
400 A
M 1749-1
PHOTOCELL READOUT
M 1740
M 1746-1
DELAY AMPLIFIER
M 1689
Fig. 2S — Development transistor— circuit packages.
* The Auto-Assembly Process used in the construction of these packages is a
Signal Corps Development .
438
THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
Fig. 29— Bit register package.
probably not use such clear plastics and Fig. 31 shows some packages
in which the plastic has been loaded with silica to increase its strength
and thermal conductivity. The assembly in Fig. 31 consists of a six-digit
position encoder at the left, followed by six regenerative pulse amplifiers
which in turn feed a six-digit combined translator-shift register.
N-P-N Transistor Audio Amplifier and Oscillator*
To the right in Fig. 32 is shown a transformer-coupled audio amplifier
employing two M1752 junction transistors. This amplifier has a pass
band from 100-20,000 cps and a power gain of approximately 90 db.
Its gain is relatively independent of collector voltage from 1-20 volts,
* The material of this section represents a summary of some work by Wallace
and Pietenpol described more completely in Ref. 4.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT
439
Fig. 30 — Package construction illustrated.
Fig. 31 — Laboratory model of encoder-transistor-register using transistor
packages.
440 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
Fig. 32— Packaged oscillator and amplifier using junction transistors.
only the available undistorted power output increasing as the voltage is
increased. At a collector voltage of 1.5 volts it draws a collector current
of approximately 0.5 ma per unit for a total power drain of 1.5 milli-
watts. Under these conditions it will deliver Class A power output of
about 0.7 milliwatt. The noise figure of such an amplifier has been
measured to be in the range from 10-15 db at 1000 cps depending upon
the operating biases.
To the left of Fig. 32 is shown a small transistor audio oscillator having
a single M1752 transistor, a transformer and one condenser. To see just
how little power was the minimum necessary to produce stable oscilla-
tions such an oscillator was tried at increasingly lower collector supply
voltages. It was found that stable oscillations could be maintained down
to collector supply voltages as low as 55 millivolts and collector current
as low as 1.5 microamperes for a total drain of Q.09 microwatt.
SUMMARY
With respect to reproducibility and interchangeability, transistors
now under development appear to be the equal of commercial vacuum
tubes.
With regard to reliability, transistors apparently have longer life and
greater mechanical ruggedness to withstand shock and vibration than
most vacuum tubes. With regard to temperature effects, transistors are
inferior to tubes and present upper limits of operation are 70-80°C for
most applications. This restriction is often reduced in importance by
the lower power consumption which results in low equipment self-
heating. This, however, is the outstanding reliability defect of transistors.
PRESENT STATUS OF TRANSISTOR DEVELOPMENT 441
With regard to miniaturization, the comparison figures are so great
as to speak for themselves. Operation with a few milliwatts is* always
feasible and in some cases operation at a few microwatts is also possible.
With regard to performance range, it is believed that the above results
imply the following tentative conclusions:
In pulse systems (up to 1-2 rac repetition rates) transistors should be
considered seriously in comparison to tubes, since they provide essen-
tially equal functional performance and have marked superiority in
miniature space and power. Bear in mind that in some reliability figures
they are superior whereas in the matter of temperature dependence
they are inferior to tubes.
In CW transmission at low frequencies (<1 mc) essentially the same
conclusions are indicated, primarily because of junction transistors. In
the range from 1-100 mc, tubes are currently superior in every functional
performance figure (except perhaps noise and bandwidth) so that for
transistors to be considered for such applications, much greater premium
must be placed on miniaturization and reliability than for the first two
applications areas.
Thus, it might be assumed that, even though there are many out-
standing development problems of a circuit and device nature to be
solved, it is appropriate for circuit engineers to explore seriously the
application possibilities of transistors — not only in the hope of building
better systems, but also to influence transistor development towards
those most important systems for which their intrinsic potentialities
best fit them. It should not be inferred that all important limitations
have been eliminated — nor, on the other hand, that the full range of
performance possibilities have been explored.
If one remembers the history of engineering research and development
in older related fields, it seems apparent that a relatively short time has
elapsed since the invention of the first point-contact transistor. Already,
new properties and new types of devices are under study and some have
been achieved in the laboratory. It therefore is possible, and certainly
stimulating, to infer that more than a single new component is involved;
that much more lies ahead than in the past; that, indeed we may be
entering a new field of technology, i.e., "transistor electronics".
ACKNOWLEDGMENTS
It was stated earlier that these advances in the development of tran-
sistors have resulted from improved understanding, materials and proces-
ses. These improvements have been made through the efforts of a large
442 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952
number of workers in physical research, chemical and metallurgical
research and transistor development. In reality, these colleagues are
the authors of this paper; and it is to them the writer owes full and
appreciative credit for the material that has made possible this report
of progress in transistor electronics.
REFERENCES
1. R. M. Ryder, R. J. Kircher, "Some Circuit Aspects of the Transistor", Bell
System Tech. J., 28, p. 367, 1949. .
2 R L. Wallace, G. Raisbeck, "Duality as a Guide in Transistor Circuit Design",
Bell Si/stem Tech. J., 30, p. 381, 1951.
3. W. Shockley, M. Sparks, G. K. Teal, "p-n Transistors", Phys. Rev., 83, p. 151,
1951. ,. .
4. R. L. Wallace, W. J. Pietenpol, "Some Circuit Properties and Applications of
n-p-n Transistors", Bell System Tech. J., 30, p. 530, 1951.
5. W. J. Pietenpol, "p-n Junction Rectifier and Photocell", Phy8. Rev., 82, No. 1,
pp. 122-121, Apr. 1, 1951.
6 J. N. Shine, "The Phototransistor", Bell Laboratories Record, 28, No. 8, pp.
337-342, 1950.